Polimeme

What this is

News aggregators tell you what happened. Polimeme identifies what is being argued about what happened, and by whom. Every day it reads the opinion pages of the press, with a focus on Europe but not only, each in the language it was written in, and assembles the arguments into discussions: the same dispute seen at once from Madrid, Warsaw, Rome, Berlin and New York.

Why it exists

Every press reads itself first — even when covering the same story. An argument born in a German column can take weeks to reach the Italian debate, laundered of its origin; an American argument may never reach a European paper at all, or arrive distorted. Polimeme makes that journey visible: where an argument started, which analysts picked it up, how and in what order, and the ones who ignored it.

How it works

1. Collection: opinion sections, analysis pieces and individual writers' publications, read continuously.
2. Grouping: pieces are matched by argument, not by keyword — a Polish column and a Spanish one about the same dispute land together, even with no word in common.
3. Editorial reading: a language model labels each discussion with what is actually in contention, the fault line, and each piece's stance within it.
4. Publication: the discussions identified are shown on the homepage, and each day is frozen into a historical archive.

How to read a discussion

Every discussion opens with the Core Contention — the question in dispute — and the Argument Map: not a simple for/against, but the two, three or four real positions in play, each piece assigned to one. The Fault Line sums up where the actual split runs, and the page flags it when a debate is near-unanimous or has already converged. Two less obvious layers sit below: how it spread — which language it started in, how many hours before it jumped to the next — and how it drifted — the way the question in dispute changed shape as new voices joined, not just more of the same. Related pieces sit apart: nearby context, not the same discussion.

Global, regional, domestic

Not every dispute travels the same distance, and the site is organised by how far each one reaches. Global covers arguments whose consequences cross borders, together with national fights being argued the same way in several countries at once — a housing rule, a family reunification law, a nuclear plant extension. Regional covers what matters to the members of a bloc as members: a shared institution decides, a precedent binds them, a border is common. Domestic covers arguments that need local context to follow at all — a leadership fight inside one party, a national scandal. A discussion can move between layers as it grows, and large ones sometimes split: when a branch takes on a life of its own, both sides stay linked and the archive of each is kept whole.

The corpus today

What it is not

Polimeme is not a news aggregator or a feed: it doesn't republish articles, it reads them and organises the arguments into discussions. It shows not a single excerpt of the original text — not even under fair use — only what the editorial reading extracts: the claim, the position, the framing. That's why it doesn't substitute for reading the originals: it never shows enough to, on purpose. Every piece referenced always links back to the source, in every discussion.

Polimeme is also not an AI crawler. Collection serves only the product itself — internal reading, to produce the analysis and the discussions — and has no other effect: it doesn't publish or republish what it collects, doesn't use it to train models, doesn't exploit it under any other pretext.

For publishers and authors

Polimeme reads your opinion and analysis, it doesn't republish it. What's visible is always derived — the claim, the position, the framing — never the original text, not even an excerpt of it. It doesn't train models on what it collects, or use it for any other purpose; every piece referenced always links back to the source.

It's a small project, still under construction, without the financial capacity for licensing deals at this point — but open to partnerships and to licensing specific to this kind of analysis, whenever and wherever that kind of programme exists. Publishers and authors who want to talk about this can write to paulo@polimeme.com.

Privacy

No advertising or third-party analytics. Polimeme keeps limited first-party server logs for security and aggregate traffic measurement; account, newsletter and billing data are stored only when those features are used. Read the full privacy notice.